Bacterial Factors That Predict Relapse after Antituberculosis Therapy

Published - Written by Amanda Fernandes, MD

Raj is a 56-year-old male who is seen in clinic for follow-up after initiating treatment for pulmonary tuberculosis (TB). He was started on standard four-drug therapy with rifampin, isoniazid, pyrazinamide, and ethambutol (RIPE). He reports gastrointestinal side effects and is finding the multidrug regimen difficult. He asks if there is a shorter regimen that would cure his TB?

Approximately 10 million cases of TB are reported worldwide per year. TB remains a public health burden given its lengthy treatment duration and associated risk of nonadherence and antimicrobial resistance. The American Thoracic Society, Centers for Disease Control and Prevention, and Infectious Diseases Society of America recommend a 6-month treatment regimen for most patients, which is associated with a 5% to 7% risk of relapse. Numerous studies have evaluated shorter treatment durations, but these regimens have been associated with relapse rates of about 20%. The strongest predictors of relapse are cavitation on initial chest radiograph and a positive acid-fast bacilli culture that persists after 2 months of intensive therapy. Could the strain of TB and its susceptibility to antibiotics also be used as a risk factor for relapse? In this week’s NEJM, Colangeli et al. address this question using data from the Tuberculosis Trials Consortium Study 22.

The Study

Study 22 was a double-blind clinical trial conducted in the U.S. and Canada in which 1004 drug-susceptible TB patients who had received 8 weeks of RIPE were randomized to receive either rifapentine and isoniazid once a week or rifampin and isoniazid twice a week for 4 months. Colangeli et al. used the data from this cohort to determine whether the baseline antimicrobial susceptibility of the infecting TB organism, based on the minimum inhibitory concentration (MIC), can be used to improve risk stratification in patients receiving treatment for TB. The MIC is the lowest concentration of an antibiotic that inhibits the growth of bacteria. Specifically, a strain of TB is considered susceptible if the MIC value is below the standard cutoff point. The authors hypothesized that the closer the MIC values were to the resistance cutoff, the less susceptible the organism is to the antimicrobial, and the greater the risk of treatment failure.

In Study 22, 57 of 1004 patients had a relapse. Pretreatment cultures were obtained from relapse patients and randomly selected patients who had been cured (controls). Factors associated with increased risk of relapse were positive culture at 8 weeks, being underweight, high MIC of isoniazid or rifampin, cavitation on imaging, white race, and bilateral disease. The investigators also found that that the MIC of samples from cured patients was lower for isoniazid and rifampin than the MIC of samples from relapse patients. The receiver operator characteristic curve with the best prediction rate for relapse was the model that combined MIC values for isoniazid and rifampin along with patient factors with an area under the curve (AUC) of 0.875 (P<0.001).

These data were combined to develop a predictive model that was then tested on a validation cohort from the National Institute of Allergy and Infectious Diseases DMID 01-009 study involving 394 patients who were recruited from 2002 through 2008 in Brazil, the Philippines, and Uganda and received TB treatment for either 4 or 6 months, depending on culture results at 2 months. The model based on MIC values for rifampin and isoniazid predicted risk of relapse, with an AUC value of 0.964. When clinical and laboratory data were added, the AUC was 0.929 (95% CI, 0.830-1.00).

The Take Away

This study suggests that checking Raj’s MIC on the pretreatment culture may be as helpful as clinical and laboratory data to estimate relapse risk and thus help determine how long he needs to be treated. In an accompanying editorial, NEJM Editor Dr. Eric Rubin concludes that MIC values “can be thought of more as probabilities of successful therapy than as absolute thresholds.”

Browse more From Pages to Practice »

 Amanda Fernandes is a 2018-2019 NEJM editorial fellow.